Welcome to Acta Armamentarii ! Today is Share:

Acta Armamentarii ›› 2022, Vol. 43 ›› Issue (6): 1415-1425.doi: 10.12382/bgxb.2021.0339

• Paper • Previous Articles     Next Articles

Optimization Method of Unmanned Swarm Defensive Combat Scheme Based on Intelligent Algorithm

MA Ye1, FAN Wenhui2, CHANG Tianqing1,2   

  1. (1.Department of Weapons and Control,Academy of Armored Force Engineering,Beijing 100072,China;2.Department of Automation,Tsinghua University,Beijing 100084,China)
  • Online:2022-03-23

Abstract: Troop deployment and task allocation are important processes of unmanned swarm defense operations, it's very important to make effective use of the limited forces in the swarm and wield the highest operational efficiency to improve the battle victory rate of unmanned swarm. At the same time, efficient combat task allocation can coordinate the consistency of swarm and better complete combat tasks. Aiming at the key combat plan in unmanned swarm defense operations, the optimization of troop deployment and coordinated task allocation in unmanned swarm defense operations is studied. A multi-agent-based unmanned swarm defensive combat model is built to quantify the key parameters required for the deployment of unmanned swarm forces, the model plans the combat area and troops, and designs the objective function. An adaptive genetic algorithm is proposed to solve the deployment problem of unmanned swarms. The proposed algorithm could dynamically adjust the objective function,crossover rate and mutation rate according to the real-time operating conditions,ensuring the inheritance of individuals with higher fitness values and avoiding the local optimization of the algorithm. A defensive operation is simulated to verify the deployment effectiveness of unmanned swarm forces. The improved deep reinforcement learning algorithm based on deep Q network is proposed to find a solution to the task allocationfor the deployed unmanned swarms. The proposed algorithm could adjust the Q value through self-adaption to avoid non-convergence caused by the algorithm's overestimation and find the optimal solution.The experimental results of defensive operations show that the proposed unmanned swarm force deployment and coordinated task allocation method could effectively improve the success rate of defensive operations,and realize the autonomous coordination and intelligent confrontation of unmanned swarms.

Key words: unmannedswarmdefenseoperations, adaptivegeneticalgorithm, agenttechnology, deepreinforcementlearning

CLC Number: